Patrick A. Singleton and Lilly Y.W. Bourguignon*

Experimental Cell Research 295 (2004) 102 118 www.elsevier.com/locate/yexcr CD44 interaction with ankyrin and IP 3 receptor in lipid rafts promotes hyaluronan-mediated Ca 2+ signaling leading to nitric oxide production and endothelial cell adhesion and proliferation Patrick A. Singleton and Lilly Y.W. Bourguignon* Endocrine Unit, Department of Medicine, University of California-San Francisco and VA Medical Center, San Francisco, CA 94121, USA Received 11 September 2003, revised version received 20 December 2003 Abstract In this study, we have showed that aortic endothelial cells (GM7372A cell line) express CD44v10 [a hyaluronan (HA) receptor], which is significantly enriched in cholesterol-containing lipid rafts (characterized as caveolin-rich plasma membrane microdomains). HA binding to CD44v10 promotes recruitment of the cytoskeletal protein, ankyrin and inositol 1,4,5-triphosphate (IP 3 ) receptor into cholesterol-containing lipid rafts. The ankyrin repeat domain (ARD) of ankyrin is responsible for binding IP 3 receptor to CD44v10 at lipid rafts and subsequently triggering HA/CD44v10-mediated intracellular calcium (Ca 2+ ) mobilization leading to a variety of endothelial cell functions such as nitric oxide (NO) production, cell adhesion and proliferation. Further analyses indicate (i) disruption of lipid rafts by depleting cholesterol from the membranes of GM7372A cells (using methyl-hcyclodextrin treatment) or (ii) interference of endogenous ankyrin binding to CD44 and IP 3 receptor using overexpression of ARD fragments (by transfecting cells with ARDcDNA) not only abolishes ankyrin/ip 3 receptor accumulation into CD44v10/cholesterol-containing lipid rafts, but also blocks HA-mediated Ca 2+ signaling and endothelial cell functions. Taken together, our findings suggest that CD44v10 interaction with ankyrin and IP 3 receptor in cholesterol-containing lipid rafts plays an important role in regulating HA-mediated Ca 2+ signaling and endothelial cell functions such as NO production, cell adhesion and proliferation. D 2004 Elsevier Inc. All rights reserved. Keywords: CD44; Ankyrin; IP 3 receptor; Lipid rafts; Ca 2+ signaling; Hyaluronan; Nitric oxide (NO) production; Endothelial cell adhesion/proliferation Introduction CD44 denotes a family of cell surface hyaluronan (HA)-binding receptors expressed in a variety of cell types [1]. Several CD44 isoforms exist because of alternative splicing of mrna encoded from a single gene [2]. Aortic endothelial cells (GM7372A cell line) have been shown to express CD44v10 which displays HA binding properties [3 5] and participates in many cellular functions [6 9]. However, the underlying mechanisms by which HA-CD44 signaling regulates these various functions are only beginning to be elucidated. Lipid rafts are a specialized plasma membrane microdomain that contains scaffolding proteins such as caveolin * Corresponding author. Endocrine Unit (111N), Department of Medicine, University of California-San Francisco and VA Medical Center, 4150 Clement Street, San Francisco, CA 94121. Fax: +1-415-383-1638. E-mail address: lillyb@itsa.ucsf.edu (L.Y.W. Bourguignon). [10,11]. Lipid components such as cholesterol, sphingomyelin and gangliosides are also present in caveolin-containing lipid rafts (also called caveolae) [10]. Caveolae are a specific subset of lipid rafts and are known to play an important role in membrane cytoskeleton interaction and calcium (Ca 2+ ) signaling [12 14]. Acylated proteins, such as GTP-binding proteins, src, fyn, hck, lck, e-nos and caveolin, are preferentially concentrated in lipid rafts [11]. In fact, acylation is thought to be one of the important mechanisms for recruiting signaling molecules into the cholesterol-containing lipid rafts [15,16] or bringing regulatory proteins to the proximal region of lipid rafts [17]. The fact that CD44 is fatty acylated [18] and that up to 40% of cellular CD44 is localized in the lipid rafts [19] suggest that acylation is critical for promoting CD44 accumulation into lipid rafts [16]. At present, the precise role of CD44 in lipid rafts during HA-mediated cellular signaling remains unclear. CD44 interacts with a family of membrane-associated cytoskeletal proteins, such as ankyrin, which are expressed 0014-4827/$ - see front matter D 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.yexcr.2003.12.025

P.A. Singleton, L.Y.W. Bourguignon / Experimental Cell Research 295 (2004) 102 118 103 in a variety of biological systems [20,21]. Ankyrin contains three functional domains: a conserved N-terminal ankyrin repeat domain (ARD) (consisting of 22 24 tandem repeats of 33 amino acids), a spectrin binding domain and a variably sized C-terminal regulatory domain [22]. CD44 binds directly to the ARD of ankyrin through a conserved ankyrinbinding domain in the CD44 cytoplasmic region. This CD44 ankyrin interaction is important for cytoskeleton activation and HA-mediated functions such as cell adhesion, proliferation and migration [20,21,23,]. Ankyrin also binds to intracellular Ca 2+ channels such as the inositol 1,4,5-trisphosphate (IP 3 ) receptor. Most importantly, ankyrin binding to IP 3 receptor modulates IP 3 binding and Ca 2+ activity [24 26]. In addition, ankyrin is involved in regulating the subcellular location of IP 3 receptors [27]. In particular, the C-terminus of ankyrin-b has been implicated in the proper subcellular localization of IP 3 receptors [27,28]. For example, the expression and subcellular distribution of IP 3 receptor are significantly altered in cardiomyocytes and thymic lymphocytes isolated from ankyrin-b knockout mice [27 29]. The physical association between IP 3 receptors and ankyrin and the dysfunction of IP 3 receptor in ankyrin- B knockout mice suggests that the IP 3 receptor and ankyrin are both structurally linked and functionally coupled. Our previous studies have indicated that HA induces CD44 ankyrin interaction [3,25] and IP 3 receptor-dependent Ca 2+ mobilization [5] leading to several important endothelial cell functions. In this paper, we have focused on the interaction of CD44 with ankyrin and IP 3 receptors in lipid rafts in endothelial cells (GM7372A). In addition, we have investigated the involvement of these interactions in regulating HA-mediated Ca 2+ signaling and endothelial cell function. Materials and methods Cell culture The chemically transformed bovine aortic cells (GM7372A) obtained from Institute for Medical Research (Camden, NJ) were grown in Eagle s minimum essential medium (EMEM) supplemented with Earle s salt solution, essential and nonessential amino acids, vitamins and 10% fetal bovine serum. Antibodies and reagents Monoclonal rat antihuman CD44 antibody (clone: 020; isotype: IgG 2b ; obtained from CMB-TECH, Inc., San Francisco, CA) used in this study recognizes a common determinant of the CD44 class of glycoproteins including CD44v10. Various antibodies such as rabbit anti- CD44v10, rabbit anti-ip 3 receptor and mouse monoclonal ankyrin antibody were prepared as described previously [5,25]. Mouse monoclonal caveolin-1 antibody was obtained form BD Transduction Laboratories. Rooster comb hyaluronan (HA) and methyl-h-cyclodextrin were purchased from Sigma Co. Cloning, expression and purification of FLAG-tagged CD44 cytoplasmic domain (CD44cyt) from E. coli The procedure for preparing the fusion protein of the FLAG-tagged cytoplasmic domain of human CD44 (FLAG-CD44cyt) was the same as described previously [30]. Cloning, expression and purification of GST-tagged ankyrin repeat domain (GST-ARD) of ankyrin The procedure for preparing the fusion protein of the GST-tagged ankyrin was the same as described previously [21]. Cell transfection For protein expression in eukaryotic cells, the cdna encoding the full-length ankyrin repeat domain (ARD) was double digested with BamHI and EcoRI and ligated in-frame into pcdna3.1/hisc vector which contains His epitope. To establish a transient expression system, cells were transfected with various plasmid DNAs (e.g. His-ARDcDNA or vector alone) using GenePORTERk methods according to the protocol provided by the manufacturer (Gene Therapy Systems, INC., San Diego, CA). Briefly, GM7372A cells were plated at a density of 2 10 6 cells per 100-mm dish and transfected with 5 or 10 Ag/dish plasmid cdna mixed with the same volume of GenePORTERk. Transfected cells were grown in the culture medium for 24 48 h. Various transfectants were then analyzed for their protein expression by immunoblot with anti-his antibody (Invitrogene). Isolation of lipid rafts and cholesterol measurement GM7372A cell monolayers (untransfected, ARDcDNAtransfected or vector-transfected cells treated with no HA or with 50 Ag/ml HA) were washed twice with PBS (ph 7.4). In some cases, cells (untransfected, ARDcDNA-transfected or vector-transfected cells) were incubated with a cholesterol depletion agent, methyl-h-cyclodextrin (5 mm), for 30 min before the treatment with 50 Ag/ml HA. These cells were then scrapped in PBS, spun down at 2000 rpm at 4jC and lysed with 0.2 ml of TN solution [25 mm Tris HCl (ph 7.5), 150 mm NaCl, 1 mm DTT, cocktail of protease inhibitors, 10% sucrose and 1% Triton X-100] for 30 min on ice. Triton X-100-insoluble materials were then mixed with 0.6 ml of cold 60% Optiprepk and overlaid with 0.6 ml of 40%, 35%, 30% and 20% Optiprepk in TN solution. The gradients were centrifuged at 35,000 rpm in SW60 rotor

104 P.A. Singleton, L.Y.W. Bourguignon / Experimental Cell Research 295 (2004) 102 118 for 12 h at 4jC and different fractions were collected. Cellular proteins or lipids associated with each fraction were precipitated according to the procedures described previously [61] and analyzed by SDS-PAGE plus immunoblotting and/or immunoprecipitation. In some cases, cholesterol content associated with different fractions was also measured using Amplex Redk cholesterol assay kit (Molecular Probes). Immunoblotting and immunoprecipitation techniques Cellular materials associated with 60%, 40%, 35%, 30% and 20% Optiprepk fractions from GM7372A cells [untreated or treated with methyl-h-cyclodextrin (5 mm); or transfected with ARDcDNA (or vector alone) in the presence or absence of 50 Ag/ml HA] were incubated with a solution containing 50 mm HEPES (ph 7.5), 150 mm NaCl, 20 mm MgCl 2, 1.0% Nonidet P-40 (NP-40), 0.2 mm Na 3 VO 4, 0.2 mm phenylmethylsulfonyl fluoride, 10 Ag/ml leupeptin and 5 Ag/ml aprotinin. The sample was then immunoprecipitated with anti-cd44v10 IgG followed by SDS-PAGE in a 5% or 7.5% polyacrylamide gel. Separated polypeptides were then transferred onto nitrocellulose filters. After blocking nonspecific sites with 2% bovine serum albumin, the nitrocellulose filters were incubated with each of the specific immunoreagents [e.g. rat anti-cd44 IgG (5 Ag/ml), rabbit anti- CD44v10 IgG (5 Ag/ml), rabbit anti-ip 3 receptor IgG (5 Ag/ml)], monoclonal antibodies against ankyrin (5 Ag/ml) and mouse monoclonal caveolin-1 antibody (5 Ag/ml) followed by incubating with horseradish peroxidase (HRP)-labeled goat anti-rat IgG, or HRP-labeled goat anti-rabbit IgG or HRP-labeled goat anti-mouse IgG or ExtrAvidin peroxidase (to detect surface-biotinylated proteins). The blots were then developed by the ECLk system (Amersham Co.). Double immunofluorescence staining GM7372A cells were first washed with PBS [0.1 M phosphate buffer (ph 7.5) and 150 mm NaCl] buffer and treated with 50 Ag/ml HA for 5 min at room temperature. In some cases, cells were not treated with HA. Untreated or HA-treated cells were then fixed by 2% paraformaldehyde and stained with Texas Red-labeled rabbit anti-cd44v10 IgG. Texas Red-labeled cells were then rendered permeable by ethanol treatment followed by incubating with fluorescein (FITC)-conjugated anticaveolin-1 IgG. To detect nonspecific antibody binding, Texas Red-anti-CD44v10- or FITC-anti-caveolin-1-labeled cells were incubated with FITC-conjugated normal rabbit serum or Texas Red-conjugated normal mouse IgG. No labeling was observed in such control samples. The fluorescein- and Texas Red-labeled samples were examined with a Leica TCS NT confocal laser scanning microscope (Leitz). In vitro binding of ARD to CD44cyt or IP 3 receptor Sepharose beads conjugated with monoclonal mouse anti-ip 3 receptor (IPR.1; it recognizes the C-terminal cytoplasmic domain of all three IP 3 receptor subtypes type I, II or III IP 3 receptors) were used to isolate IP 3 receptors as described previously [24,25]. Purified IP 3 receptor subtypes contain intact IP 3 receptor type I (IP 3 RI), IP 3 receptor type II (IP 3 RII) and IP 3 receptor type III (IP 3 RIII) (with molecular mass of approximately 260 kda) as detected by anti-ip 3 RI, anti-ip 3 RII or anti-ip 3 RIII immunoblot, respectively. Aliquots (0.5 1 ng protein) of purified IP 3 receptor-conjugated Sepharose beads were incubated in 0.5 ml of binding buffer [20 mm Tris HCl (ph 7.4), 150 mm NaCl, 0.1% bovine serum albumin and 0.05% Triton X-100] containing various concentrations (10 800 ng/ml) of 125 I-labeled the ARD fusion protein (5000 cpm/ng protein) at 4jC for 4 h. Following binding, the IP 3 receptor-conjugated beads were washed extensively in binding buffer and the beads-bound radioactivity was counted. Nonspecific binding was determined using a 50- to 100-fold excess of unlabeled ARD in the presence of the same concentration of 125 I-labeled ARD. In some cases, we have incubated ARD-conjugated beads with biotinylated CD44cyt in the presence of various concentrations of unlabeled CD44cyt ranging from 10 12 to 10 6 M. Following the binding, biotinylated CD44cyt associated with ARD-conjugated beads was detected by ExtrAvidin peroxidase reaction. Nonspecific binding, which was approximately 20% of the total binding, was always subtracted from the total binding. Reconstitution of IP 3 R in phospholipid vesicles (liposomes) and measurement of 45 Ca 2+ flux IP 3 receptor isolated from lipid rafts (using vectortransfected cells or ARDcDNA-transfected cells treated with HA or no HA) was used for the measurement of IP 3 -induced Ca 2+ influx. The Ca 2+ flux measurement was initiated by adding 2 ACi of 45 Ca 2+ (ICN Pharmaceuticals Inc., Costa Mesa, CA) to the IP 3 receptor-containing liposomes in 20 mm Tris HCl (ph 7.4), 100 mm NaCl, 100 mm KCl, in the presence or absence of IP 3 (1 AM) at 30jC, in a final volume of 50 Al. The Ca 2+ flux measurement was terminated by adding 3-fold excess of 0.5 mm CaCl 2, 5 mm MgSO 4 and 100 Ag/ml heparin, and external 45 Ca 2+ was removed by filtration using HAWP filters (0.45 Am, Millipore). After extensive wash with 20 mm Tris HCl (ph 7.4), 100 mm NaCl and 100 mm KCl, filterassociated radioactivity was determined by liquid scintillation counting. Measurement of intracellular Ca 2+ mobilization GM7372A cells (10 7 cells/ml) (e.g. untransfected or transfected with ARDcDNA or vector only) [pretreated with Xestospongin C (10 AM) or methyl-h-cyclodextrin

P.A. Singleton, L.Y.W. Bourguignon / Experimental Cell Research 295 (2004) 102 118 105 (5 mm) or without treatment] were first incubated with 10 AM Fura-2/AM (Calbiochem) for 1 h at room temperature in a buffer solution containing 145 mm NaCl, 5 mm KCl, 0.1 mm MgCl 2, 5 mm glucose and 15 mm HEPES (ph 7.3) in the presence or absence of 1 mm CaCl 2. Cells were subsequently washed three times with the same buffer. Cells (10 7 cells/ml) resuspended in 0.1 M phosphate saline buffer (ph 7.0) were incubated simultaneously with an equal volume of 0.1 M phosphate saline buffer (ph 7.0) containing HA (50 Ag/ml) into a 20-Am chamber alternately illuminated with 200-ms flashes of 340 and 380 nm every 10 ms (monitoring the emission wavelength of 510 nm) using a Dural-wavelength Fluorescence Imaging System (Intracellular Imaging Inc., Cincinnati, OH). The concentration of intracellular Ca 2+ was determined by the following equation: Ca 2+ = K d [(R R min ) / (R max R)]F/B, where Ca 2+ is intracellular Ca 2+, K d is the dissociation constant of Fura-2 for Ca 2+, R is the ratio of the Fura-2 fluorescence excited at 340 divided by the fluorescence excited at 380 nm, R min and R max minimal and maximal fluorescence ratios, respectively, obtained in ionomycin in the presence of 7 mm EGTA or 2.0 mm Ca 2+. F and B are the fluorescence voltage signals at 380 nm in 5.0 AM ionomycin in the presence of 7 mm EGTA and 2 mm Ca 2+, respectively. Measurement of nitric oxide (NO) production GM7372A cells [approximately 1 10 4 cells/well in phosphate-buffered saline (PBS), ph 7.2, pretreated with various agents, e.g., anti-cd44 IgG, (1 Ag/ml) methyl-hcyclodextrin (5 mm), Xestospongin C (10 AM), BAPTA/AM (1 AM) for 1 h or transfected with ARDcDNA (or vector alone)] were incubated with HA (50 Ag/ml) for 30 min followed by nitric oxide (NO) production measurement using a nitric oxide (NO) quantitation kit (Active Motif, Carlsbad, CA). Each assay were set up in triplicate, repeated at least five times and analyzed statistically by Student s t test (with statistical significance set at P < 0.01). Cell adhesion and proliferation assays For cell adhesion assay, GM7372A cells [5 10 3 cells/ well pretreated with various agents (e.g. anti-cd44 IgG, (1 Ag/ml), methyl-h-cyclodextrin (5 mm), Xestospongin C (10 AM), BAPTA/AM (1 AM) for 1 h or transfected with ARDcDNA (or vector alone)] were incubated with 0.2 ml of Dulbecco s modified Eagle s medium in 96-well culture plates coated with HA or no HA. Cell associated with these plates was detected using the MTT [3-(4,5-dimethylthiazol- 2-yl)-2,5-diphenyltetrazolium bromide] assays. For measuring cell growth, GM7372A cells [5 10 3 cells/well pretreated with various agents (e.g. anti-cd44 IgG, (1 Ag/ml), methyl-h-cyclodextrin (5 mm), Xestospongin C (10 AM), BAPTA/AM (1 AM) for 1 h or transfected with ARDcDNA (or vector alone)] were incubated with 0.2 ml of EMEM containing either HA (50 Ag/ml) (or no HA) for 24 h at 37jC in 5%CO 2 /95% air in 96-well culture plates. The in vitro cell proliferation assay was analyzed by measuring increases in cell number using the MTT assays. Each assay were set up in triplicate, repeated at least five times and analyzed statistically by Student s t test (with statistical significance set at P < 0.01). Results Analysis of CD44v10-containing lipid rafts in endothelial cells As with many other cell types, endothelial cells contain cholesterol-rich lipid raft microdomains. One type of lipid rafts is the caveolae, which contains a specific scaffolding protein called caveolin [10,11]. Our previous studies have shown that a CD44 isoform, such as CD44v10, functions as a major hyaluronan (HA) receptor in aortic endothelial cells (GM7372A) [3 5]. Using double immunofluorescence staining and confocal microscopic analyses, we have determined that CD44v10 is expressed on the cell surface (Fig. 1A-a) and caveolin-1 is diffusely distributed in the cytoplasm of untreated GM7372A cells (Fig. 1A-b). CD44v10 caveolin-1 co-localization is not significant in these cells (Fig. 1A-c). Upon HA addition, there is a dramatic redistribution of caveolin-1 from the cytosol to the plasma membranes (Fig. 1B-b) plus co-localization of CD44v10 (Fig. 1B-a) and caveolin-1 (Fig. 1B-b) into discrete patches along the plasma membrane of GM7372A cells (Fig. 1B-c). These results suggest that HA can induce a close association between CD44 and caveolin-containing microdomain in the aortic endothelial GM7372A cells. In recent years, many researchers have used a Triton X- 100 solubility technique (to isolate Triton X-100-insoluble materials) followed by flotation experiments in Optiprepk gradient centrifugation to biochemically identify lipid rafts or caveolin-containing microdomains [14,31]. To test whether CD44v10 is associated with lipid rafts or caveolin-containing microdomains during HA signaling, we have isolated Triton X-100-insoluble materials (obtained from GM7372A cells with or without HA treatment) which were then loaded onto a 60% Optiprepk gradient layer. Following flotation centrifugation, we have observed that some caveolin-1 is able to float from the 60% Optiprepk layer to the 20 30% Optiprepk layers (Fig. 2A-I-a and II-a). Our results agree with previous findings showing that caveolin-1 is present in the 20 30% Optiprepk fraction. Furthermore, it is noted that a small amount of CD44v10 (approximately 10 15% of CD44v10) is also partitioned at the 20 35% Optiprepk layers using Triton X-100-insoluble materials isolated from untreated GM 7372A cells (Fig. 2A-I-b). Interestingly, a significant amount of CD44v10 (approximately 30 40% of CD44v10) is recruited into the 20 30%

106 P.A. Singleton, L.Y.W. Bourguignon / Experimental Cell Research 295 (2004) 102 118 Fig. 1. Double immunofluorescence staining of CD44v10 and caveolin-1 in GM7372A cells. Aortic endothelial cells (GM7372A cell line) were treated with 50 Ag/ml HA for 5 min at room temperature, fixed by 2% paraformaldehyde and surface labeled with Texas Red-labeled anti-cd44v10 antibody (a). These cells were then rendered permeable by ethanol treatment and stained with FITC-labeled anti-caveolin-1 antibody (b). (A) Texas Red-conjugated anti-cd44v10 (a), FITC-conjugated anti-caveolin-1 (b) and co-localization of CD44v10 and caveolin-1 (c) in untreated cells. (B) Texas Red-conjugated anti-cd44v10 (a), FITCconjugated anti-caveolin-1 (b) and co-localization of CD44v10 and caveolin-1 (c) in HA-treated cells. Optiprepk fractions from the 60% Optiprepk layer using Triton X-100-insoluble materials obtained from GM7372A cells treated with HA (Fig. 2A-II-b). Since Triton X-100- insoluble materials floating to the low-density region (i.e. 20 30% layers) represent the lipid rafts, it appears that HA promotes CD44v10 accumulation into caveolin-1-containing microdomain such as lipid rafts in aortic endothelial GM7372A cells. Cholesterol has been shown to be important in lipid raft formation and cellular signaling [32]. In this study, we have Fig. 2. Detection of CD44v10 in caveolin-1 and cholesterol-containing lipid rafts isolated from GM7372A cells. Triton X-100-insoluble material isolated from GM7372A cells was mixed with 0.6 ml of cold 60% Optiprepk and overlaid with 0.6 ml of 40%, 35%, 30% and 20% Optiprepk. The gradients were centrifuged at 35,000 rpm in SW60 rotor for 12 h at 4jC and different fractions were collected. The 20 30% Optiprepk layer represents the lipid rafts. Cellular protein or cholesterol-associated with each fraction was analyzed as described in Materials and methods. (A-I) Detection of cellular protein associated with 20 60% Optiprepk layers (isolated from untreated cells) using anti-caveolin-1 (a) or anti-cd44v10 (b)-mediated immunoblotting analyses. (A-II) Detection of cellular protein associated with 20 60% Optiprepk layers (isolated from HA treated cells) using anti-caveolin-1 (a) or anti-cd44v10 (b)- mediated immunoblotting analyses. (A-III) Detection of cellular protein associated with 20 60% Optiprepk layers (isolated from cells pretreated with a cholesterol depletion agent, methyl-h-cyclodextrin followed by HA treatment) using anti-caveolin-1 (a) or anti-cd44v10 (b)-mediated immunoblotting analyses. (B) Measurement of cholesterol content in lipid rafts (the 20 30% Optiprepk layer) isolated from untreated cells (a) or methyl-h-cyclodextrintreated cells (b).

P.A. Singleton, L.Y.W. Bourguignon / Experimental Cell Research 295 (2004) 102 118 107 measured the cholesterol content associated with lipid rafts (the 20 30% Optiprepk fractions) (Fig. 2B) isolated from GM7372A cells. Our results indicate that approximately 13 Ag cholesterol/mg total protein is present in the lipid raft isolated from GM7372A cells (Fig. 2B-a). Treatment of GM7372A cells with methyl-h-cyclodextrin (an agent known to disrupt lipid rafts by depleting cholesterol from the membranes) reduces the cholesterol content (approximately 2 Ag/mg protein) of lipid rafts (Fig. 2B-b) in GM7372A cells. Following methyl-h-cyclodextrin treatment, there is a redistribution of caveolin-1 away from lipid raft fractions (Fig. 2A-III-a). Specifically, most of CD44v10 remains at 60% Optiprepk and only a small amount of CD44v10 floats at 40% Optiprepk layer using Triton X- 100-insoluble materials isolated from GM7372 cells pretreated with methyl-h-cyclodextrin followed by HA addition (Fig. 2A-III-b). The failure of CD44v10 and caveolin-1 to partition into lipid rafts (the 20 30% Optiprepk fractions) after the removal of cholesterol from the membrane suggests that cholesterol plays an important role in stabilizing the CD44v10 association with caveolin-1-containing lipid rafts. Detection of HA-induced recruitment of ankyrin and the IP 3 receptor to CD44v10-containing lipid rafts Lipid rafts are known to participate in membrane cytoskeleton interactions during signal transduction [13,14]. In this study, we have investigated whether ankyrin and the IP 3 receptors are also present in CD44v10-associated lipid rafts (Fig. 3). Using antiankyrin and anti-ip 3 receptor-mediated immunoblot analyses, we have detected both ankyrin (approximately 216 kda protein) (Fig. 3A-I-b and A-II-b) and IP 3 receptor (approximately 260 kda protein) (Fig. 3A-I-c and A-II-c) are detected in the Triton X-100-insoluble materials which were loaded onto 60% Optiprepk layer. After flotation and immunoblotting (with anti-ankyrin, anti-ip 3 receptors or anti-cd44v10, respectively), we have found that neither ankyrin (Fig. 3A-I-b) nor the IP 3 receptors (Fig. 3A-I-c) are present in the CD44v10-associated lipid rafts (Fig. 3A-I-a) from untreated GM7372A cells. However, upon HA treatment of GM7372A cells, both ankyrin (Fig. 3A-II-b) and IP 3 receptors (Fig. 3A-II-c) together with CD44v10 (Fig. 3A-II-a) become partitioned into the 20 30% Optiprepk layer containing lipid rafts (Fig. 3). Disruption of the structural integrity of lipid rafts by depleting cholesterol (using methyl-h-cyclodextrin treatment) inhibits HA-induced recruitment of ankyrin and IP 3 receptor into CD44v10-containing lipid rafts (Fig. 3A-III-a to c). Furthermore, we have immunoprecipitated cellular materials in lipid rafts [the 20 30% Optiprepk layer isolated from untreated (Fig. 3A-I) or HA-treated cells (Fig. 3A-II)] with anti-cd44v10 antibody followed by anti-ankyrin or anti-ip 3 receptor-mediated immunoblot (or Fig. 3. Detection of ankyrin and IP 3 receptor in CD44v10-containing lipid rafts isolated from GM7372A cells. Triton X-100-insoluble materials isolated from GM7372A cells was mixed with 0.6 ml of cold 60% Optiprepk and overlaid with 0.6 ml of 40%, 35%, 30% and 20% Optiprepk. The gradients were centrifuged at 35,000 rpm in SW60 rotor for 12 h at 4jC and different fractions were collected. The 20 30% Optiprepk layer represents the lipid rafts. Cellular protein or cholesterol associated with each fraction was analyzed as described in Materials and methods. (A-I) Detection of cellular protein associated with 20 60% Optiprepk layers (isolated from untreated cells) using anti-cd44v10 (a), anti-ankyrin (b) or anti-ip 3 receptor (c)-mediated immunoblotting analyses. (A-II) Detection of cellular protein associated with 20 60% Optiprepk layers (isolated from HA treated cells) using anti-cd44v10 (a), anti-ankyrin (b) or anti-ip 3 receptor (c)-mediated immunoblotting analyses. (A-III) Detection of cellular protein associated with 20 60% Optiprepk layers (isolated from cells pretreated with a cholesterol depletion agent, methyl-hcyclodextrin followed by HA treatment) using anti-cd44v10 (a), antiankyrin (b) or anti-ip 3 receptor (c)-mediated immunoblotting analyses. (B- I) Analysis of the complex formation between ankyrin/ip 3 receptor and CD44v10 using anti-cd44v10-mediated immunoprecipitation of cellular material-associated with lipid rafts (isolated from untreated cells) followed by immunoblotting with anti-ankyrin antibody (a) or anti-ip 3 receptor antibody (b); or reblotting with anti-cd44v10 antibody (c). (B-II) Analysis of the complex formation between ankyrin/ip 3 receptor and CD44v10 using anti-cd44v10-mediated immunoprecipitation of cellular material-associated with lipid rafts (isolated from HA-treated cells) followed by immunoblotting with anti-ankyrin antibody (a) or anti-ip 3 receptor antibody (b); or reblotting with anti-cd44v10 antibody (c). reblotting with anti-cd44v10). Our results indicate that very little ankyrin (Fig. 3B-I-a) or IP 3 receptors (Fig. 3B-Ib) are detected in the anti-cd44v10-mediated immunoprecipitated material (Fig. 3B-I-c) using lipid rafts from untreated cells. However, again both ankyrin (Fig. 3B-IIa) and IP 3 receptors (Fig. 3B-II-b) are found to be tightly complexed with CD44v10 (Fig. 3B-II-c) in lipid rafts

108 P.A. Singleton, L.Y.W. Bourguignon / Experimental Cell Research 295 (2004) 102 118 using GM7372A cells treated with HA. These observations suggest that HA promotes recruitment of signaling complexes, including the cytoskeletal protein (e.g. ankyrin) and Ca 2+ channels (e.g. IP 3 receptors), into lipid rafts of aortic endothelial cells. Analyses of ankyrin repeat domain (ARD) interactions with CD44v10 and IP 3 receptors Ankyrins contain three functional domains: ankyrin repeat domain (ARD), spectrin binding domain (SBD) Fig. 4. Ankyrin structure, ankyrin repeat domain (ARD) and ARD interaction with CD44v10 and IP 3 receptor in vitro. Schematic illustration of functional domains in full-length ankyrin: ankyrin repeat domain (ARD), spectrin binding domain (SBD) and regulatory domain (RD). (A-a,-b) ARDcDNA was constructed according to strategy described in Materials and methods. (B) Scatchard plot analysis of equilibrium binding between 125 I-labeled ARD and IP 3 receptor. Specifically, the IP 3 receptor-conjugated beads were incubated with various concentrations (10 800 ng/ml) of 125 I-labeled the ARD fusion protein (5000 cpm/ng protein) at 4jC for 4 h. Following binding, the IP 3 receptor-conjugated beads were washed extensively in binding buffer and the beads-bound radioactivity was counted. Nonspecific binding was determined using a 50- to 100-fold excess of unlabeled ARD in the presence of the same concentration of 125 I-labeled ARD. (C) Binding interaction between ARD and CD44cyt (the cytoplasmic domain of CD44 fusion protein). Specifically, ARD-conjugated beads were incubated with biotinylated CD44cyt in the presence of various concentrations of unlabeled CD44cyt ranging from 10 12 to 10 6 M. Following the binding, biotinylated CD44cyt associated with ARD-conjugated beads was detected by ExtrAvidin peroxidase reaction. Nonspecific binding, which was approximately 20% of the total binding, was always subtracted from the total binding.

P.A. Singleton, L.Y.W. Bourguignon / Experimental Cell Research 295 (2004) 102 118 109 and a variably sized C-terminal regulatory domain (RD) (Fig. 4A) [22]. In particular, the N-terminal region of ARD is composed of a tandem array of 24 ankyrin repeats and is known to interact with membranes (Fig. 4A-a and b). The question whether the membrane-binding domain of ankyrin (i.e. the ARD) is involved in binding to CD44v10 and/or IP 3 receptors is addressed in this study. First, a pgex-2tk recombinant plasmid encoding ARD (N-terminal portion of ankyrin, from aa 1 to 834) was constructed with a GST tag and expressed in E. coli (Fig. 4A-a) [21]. The purified GST-tagged ARD fusion protein was then tested for its ability to bind either IP 3 receptors (Fig. 4B) or the cytoplasmic domain of CD44 (CD44cyt) (Fig. 4C). First, we examined the binding of IP 3 receptors to a 125 I-labeled GST-ARD fragment under equilibrium binding conditions. Scatchard plot analysis shows that ARD fragment binds to IP 3 receptors at a single site (Fig. 4B) with high affinity [an apparent dissociation constant (K d ) of approximately 0.19 nm]. We have also incubated ARD-conjugated beads with various concentrations of biotinylated-labeled CD44 cytoplasmic domain fragment (FLAG-CD44cyt) [21,30] under equilibrium binding conditions. Our results indicate that CD44 also binds to ARD at a single site with an apparent dissociation constant (K d ) of approximately 3.5 nm (Fig. 4C). These findings strongly support the notion that ankyrin [particularly, the ankyrin repeat domain (ARD)] is involved in the binding of both CD44 and IP 3 receptors. To further analyze the interaction between the ARD fragment of ankyrin and CD44v10 and IP 3 receptors in vivo, we have constructed the ARDcDNA (Fig. 4A-b). This construct was then cloned into a pcdna3.1/hisc expression vector followed by a transient transfection of His-tagged ARDcDNA (Fig. 4A-b) (or vector alone) into GM7372A cells. By carrying out an anti-his immunoprecipitation and immunoblot (with anti-his, anti-cd44v10 or anti-ip 3 receptors) of GM7372A cells transfected with His-tagged ARDcDNA (or vector alone), we have detected the expression of the His-tagged ARD fragment of ankyrin in His-tagged ARDcDNA tranfected cells (Fig. 5A-II-c). No protein band was revealed in vector-transfected GM7372A cells using the same anti-his immunoprecipitation and immunoblot procedures (Fig. 5A-I-c). Immunoblotting of cell lysates obtained from ARDcDNA (Fig. 5B-II) or vector-transfected cells (Fig. 5B-I) with various immuno-reagents (e.g. anti-cd44v10, anti-ip 3 receptors and anti-ankyrin) shows that CD44v10 (Fig. 5B-I-a and B-II-a), IP 3 receptors (Figs. 5B-I-b and B-IIb) and ankyrin (Fig. 5B-I-c and B-II-c) are expressed at comparable levels in these transfectants. When these transfectants were immunoprecipitated by anti-his antibody followed by anti-cd44v10 or IP 3 receptors immunoblot, respectively, we found that both CD44v10 (Fig. 5A-II-a) and IP 3 receptors (Fig. 5A-II-b) are co-precipitated with the His-tagged ARD fragment of ankyrin from cells transfected with His-ARDcDNA (Fig. 5A-II-c). No detectable CD44v10 (Fig. 5A-I-a) or IP 3 receptors (Fig. 5A-I-b) are found in anti-his-mediated immunoprecipitated materials isolated from vector-transfected cells (Fig. 5A-I-c). These results confirmed that the ARD domain of ankyrin is complexed with CD44v10 and IP 3 receptors in vivo. Moreover, we have demonstrated that HA can promote the recruitment of endogenous ankyrin (Fig. 5C-I-a) and cellular IP 3 receptors (Fig. 5C-I-b) into a complex with CD44v10 (Fig. 5C-I-c) using anti- CD44v10-mediated immunoprecipation followed by antiankyrin- or anti-ip 3 receptors (or anti-cd44v10)-mediated immunoblot, respectively, in vector-transfected GM7372A cells (Fig. 5C-I). In contrast, our results show that transfection of GM7372A cells with ankyrin s ARDcDNA not only causes a large reduction in endogenous ankyrin (Fig. 5C-II-a) or IP 3 receptor (Fig. 5C-II-b) association with CD44v10 (Fig. 5C-II-c), but also causes a marked inhibition of HA-mediated recruitment of both ankyrin (Fig. 5C-II-a) and IP 3 receptors (Fig. 5C-II-b) to CD44v10 (Fig. 5C-II-c). In addition, we have analyzed the distribution of CD44v10, ankyrin and IP 3 receptors in Triton X-100- insoluble materials isolated from either vector-transfected cells (Fig. 5D-I and D-II) or ARDcDNA-transfected cells using the Optiprepk gradient centrifugation technique (Fig. 5D-III and D-IV). After flotation and immunoblotting (with anti-cd44v10, anti-ankyrin or anti-ip 3 receptor, respectively), we have determined that the amount of ankyrin (Fig. 5D-I-b) or anti-ip 3 receptor (Fig. 5D-I-c) associated with CD44v10-containing lipid rafts is very little in vector-transfected cells treated with no HA (Fig. 5D-I-a to c). However, a significant recruitment of ankyrin (Fig. 5D-II-b) and IP 3 receptors (Fig. 5D-II-c) into CD44v10 (Fig. 5D-II-a)-containing lipid rafts (the 20 30% Optiprepk layer) occurs in vector-transfected cells treated with HA (Fig. 5D-II). Although CD44v10 accumulation (Fig. 5D-III-a and D-IV-a) at lipid rafts occurs in ARDcDNA-transfected cells in the presence or absence of HA treatment, neither ankyrin (Fig. 5D-III-b and D-IV-b) nor the IP 3 receptors (Fig. 5D-III-c and D-IV-c) are partitioned into the CD44v10-associated lipid rafts in these transfectants. These results clearly support the conclusion that the ankyrin repeat domain (ARD) fragment acts as a potent competitive inhibitor for endogenous, intact ankyrin and IP 3 binding to CD44v10 in vivo; and it also functions as a strong dominant-negative mutant that blocks HA-induced ankyrin and IP 3 receptor localization into lipid rafts. Involvement of CD44v10/ankyrin-associated IP 3 receptors in regulating Ca 2+ signaling and HA-mediated endothelial cell functions HA CD44 interaction promotes an IP 3 receptor-dependent increase in intracellular calcium in aortic endothelial cells [5]. Ca 2+ signaling is considered to be very impor-

110 P.A. Singleton, L.Y.W. Bourguignon / Experimental Cell Research 295 (2004) 102 118 tant for HA/CD44-mediated endothelial cell function. Our previous studies have shown that ankyrin binding to the IP 3 receptor regulates IP 3 -binding and IP 3 -mediated Ca 2+ flux [24]. To address the question whether the IP 3 receptor recruited to CD44v10-containing lipid rafts via ankyrin is functionally responsive to IP 3 -mediated Ca 2+ fluxes, we first isolated CD44v10/ankyrin-linked IP 3 receptors from lipid rafts of vector-transfected cells [trea-

P.A. Singleton, L.Y.W. Bourguignon / Experimental Cell Research 295 (2004) 102 118 111 ted with HA (Fig. 5D-II) or no HA (Fig. 5D-I)]. Our results indicate that IP 3 receptors (complexed with CD44v10 and ankyrin) isolated from lipid rafts of vector-transfected cells (treated with HA) display a higher level of IP 3 -induced Ca 2+ flux activity (Fig. 6A-I-b) compared to the IP 3 receptor isolated from lipid rafts of vector-transfected cells treated with no HA (Fig. 6A-I-b). In contrast, the ability of IP 3 receptors isolated from lipid rafts of ARDcDNA-transfected cells, either in the presence of HA (Fig. 5D-IV) or in the absence of HA (Fig. 5D-III), to conduct IP 3 -mediated Ca 2+ flux is readily abolished. These observations suggest that ankyrin plays an important role in recruiting IP 3 receptors into CD44v10-containing lipid rafts required for HA-induced IP 3 receptor function (e.g. Ca 2+ flux). We have also used the fluorescence indicator, Fura-2, to measure intracellular free Ca 2+ concentrations after HA binding to CD44v10-containing GM7372A cells. The ratio of the fluorescence signal from Fura-2 at 340 and 380 nm excitation was used to determine the intracellular Ca 2+ concentration. Our results clearly showed that the intracellular Ca 2+ concentration is elevated shortly after the addition of HA to vector-transfected GM7372A cells, and this increase is followed by a continuous Ca 2+ influx (Fig. 6B-a). This HA-induced rise in intracellular Ca 2+ can be inhibited by pretreatment of cells with Xestospongin C (a membrane permeable blocker of IP 3 -mediated Ca 2+ release) (Fig. 6B-b) or methyl-h-cyclodextrin, an agent known to deplete cholesterol from membranes and disrupt lipid raft (Fig. 6B-c). The addition of ionomycin (5 AM) reveals that treatment of cells with Xestospongin C (Fig. 6B-b, insert) or methyl-h-cyclodextrin (Fig. 6B-c, insert) does not significantly affect internal Ca 2+ stores. Ionomycin-induced Ca 2+ elevation can be readily inhibited by EGTA (Figs. 6B-b, insert and B-c, insert). Finally, we have observed that overexpression of ARD by transfecting GM7372A cells with ARDcDNA blocks HA-mediated Ca 2+ elevation (Fig. 6B-d). The addition of ionomycin to these ARDcDNA-transfected cells also does not affect the level of Ca 2+ in the stores (Fig. 6B-d, insert). In the presence of EGTA, ionomycin-induced Ca 2+ elevation can be readily inhibited (Fig. 6B-d, insert). Taken together, these results suggest that the interaction of ankyrin with IP 3 receptor in CD44-containing lipid rafts plays an important role in HA-mediated Ca 2+ signaling in endothelial cells. Further analyses show that HA promotes several important endothelial cell functions including nitric oxide (NO) production (Table 1), endothelial cell adhesion (Table 2) and proliferation (Table 3). The fact that treatment of GM7372A cells with rat anti-cd44 antibody (but not normal rat IgG) (Table 1) or various agents, such as Xestospongin C (an IP 3 receptor inhibitor) or a membrane-permeable Ca 2+ chelator (BAPTA), blocks HA and CD44-mediated endothelial cell functions [e.g. NO production (Table 1), cell adhesion (Table 2) and proliferation (Table 3)] supports the notion that IP 3 receptor-mediated Ca 2+ signaling is required for HA/CD44-mediated endothelial cell functions. Finally, we have observed that (i) disruption of lipid rafts by depleting cholesterol from the membranes of GM7372A cells (using methyl-h-cyclodextrin treatment) or (ii) interference of endogenous ankyrin binding to CD44 and IP 3 receptor using overexpression of ARD fragments (by transfecting cells with ARDcDNA) not only abolishes ankyrin/ip 3 receptor accumulation into CD44v10/cholesterol-containing lipid rafts (Fig. 5D), but also significantly inhibits Ca 2+ Fig. 5. Analyses of ARD interaction with CD44v10 and IP 3 receptor in GM7372A cells. GM7372A cells (transfected with His-tagged ARDcDNA or vector alone) were treated with HA (or no HA). These transfectants were then solubilized by NP-40, and immunoblotted with various antibodies (e.g. anti-his, anti- CD44v10, anti-ip 3 receptor, anti-ankyrin) or immunoprecipitated with anti-his antibody/anti-cd44v10 followed by immunoblotting with various immunoreagents. In some cases, Triton X-100-insoluble material isolated from transfectants was mixed with 0.6 ml of cold 60% Optiprepk and overlaid with 0.6 ml of 40%, 35%, 30% and 20% Optiprepk. The gradients were centrifuged at 35,000 rpm in SW60 rotor for 12 h at 4jC and different fractions were collected as described in Materials and methods. (A-I) Analysis of the complex formation between CD44v10/IP 3 receptor and His-tagged ARD in vector-transfected cells using anti-his-mediated immunoprecipitation followed by immunoblotting with anti-cd44v10 antibody (a) or anti-ip 3 receptor antibody (b); or reblotting with anti-his antibody (c). (A-II) Analysis of the complex formation between CD44v10/IP 3 receptor and His-tagged ARD in ARDcDNA-transfected cells using anti-his-mediated immunoprecipitation followed by immunoblotting with anti-cd44v10 antibody (a) or anti-ip 3 receptor antibody (b); or reblotting with anti- His antibody (c). (B-I) Detection of CD44v10/ankyrin/IP 3 receptor expression in cell lysate isolated from vector-transfected cells using anti-cd44v10 (a), antiankyrin (b) or anti-ip 3 receptor-mediated immunoblotting. (B-II) Detection of CD44v10/ankyrin/IP 3 receptor expression in cell lysate isolated from ARDcDNA-transfected cells using anti-cd44v10 (a) or anti-ankyrin (b) or anti-ip 3 receptor-mediated immunoblotting. (C-I) Analysis of the complex formation between ankyrin/ip 3 receptor and CD44v10 in vector-transfected cells (treated with HA or no HA) using anti-cd44v10-mediated immunoprecipitation followed by immunoblotting with anti-ankyrin antibody (a) or anti-ip 3 receptor antibody (b); or reblotting with anti-cd44v10 antibody (c). (C-II) Analysis of the complex formation between ankyrin/ip 3 receptor and CD44v10 in vector-transfected cells (treated with HA or no HA) using anti- CD44v10-mediated immunoprecipitation followed by immunoblotting with anti-ankyrin antibody (a) or anti-ip 3 receptor antibody (b); or reblotting with anti- CD44v10 antibody (c). (D-I) Detection of cellular protein associated with 20 60% Optiprepk layers (isolated from vector-transfected cells treated with no HA) using anti-cd44v10 (a), anti-ankyrin (b) or anti-ip 3 receptor (c)-mediated immunoblotting analyses. (D-II) Detection of cellular protein associated with 20 60% Optiprepk layers (isolated from vector-transfected cells treated with HA) using anti-cd44v10 (a), anti-ankyrin (b) or anti-ip 3 receptor (c)-mediated immunoblotting analyses. (D-III) Detection of cellular protein associated with 20 60% Optiprepk layers (isolated from ARDcDNA-transfected cells treated with no HA) using anti-cd44v10 (a), anti-ankyrin (b) or anti-ip 3 receptor (c)-mediated immunoblotting analyses. (D-IV) Detection of cellular protein associated with 20 60% Optiprepk layers (isolated from ARDcDNA-transfected cells treated with HA) using anti-cd44v10 (a), anti-ankyrin (b) or anti-ip 3 receptor (c)-mediated immunoblotting analyses.

112 P.A. Singleton, L.Y.W. Bourguignon / Experimental Cell Research 295 (2004) 102 118 signaling (Fig. 6) and HA/CD44-specific endothelial cell functions [e.g. NO production (Table 1), endothelial cell adhesion (Table 2) and proliferation (Table 3)]. These findings suggest that CD44 interaction with ankyrin and IP 3 receptor in lipid rafts plays an important role in regulating both Ca 2+ signaling and Ca 2+ -depedent endothelial cell function such as NO production, cell adhesion and proliferation. Discussion Aortic endothelial cells are known to play an important role in producing potent vascular agents contributing to both angiogenesis and modulating vascular tone [33,34]. Abnormal aortic endothelial cell function is considered to be one of the key factors in promoting atherosclerosis and cardiovascular diseases [35 37]. Hyaluronan (HA), the major glycos-

P.A. Singleton, L.Y.W. Bourguignon / Experimental Cell Research 295 (2004) 102 118 113 Table 1 Measurement of HA-dependent and CD44v10-specific nitric oxide (NO) production in endothelial cells (A) Effects of anti-cd44 antibody on HA-dependent and CD44v10-specific NO production in endothelial cells Treatments NO production (nmol/mg total cellular protein) a No treatment (control) 0.42 HA treatment 4.30 Anti-CD44 IgG pretreatment 0.80 + HA treatment (B) Effects of various drugs on HA-dependent and CD44v10-specific NO production in endothelial cells Treatments NO production (nmol/mg total cellular protein) a No drug treatment 4.2 (control) + HA Methyl-h-cyclodextrin 0.5 treatment + HA Xestospongin C 0.3 treatment + HA BAPTA treatment + HA 0.1 (C) Effect of ARD overexpression on HA-dependent and CD44v10-specific NO production in endothelial cells Cells NO production (nmol/mg total cellular protein) a No HA addition HA addition Untransfected cells (control) 0.42 4.30 Vector-transfected cells 0.50 4.90 ARDcDNA-transfected cells 0.30 0.35 a GM7372A cells [approximately 1 10 4 cells/well pretreated with various agents (e.g. anti-cd44 IgG, (1 Ag/ml) methyl-h-cyclodextrin (5 mm), Xestospongin C (10 AM), BAPTA/AM (1 AM) for 1h or transfected with ARDcDNA (or vector alone)] were incubated with HA (50 Ag/ml) for 30 min followed by nitric oxide (NO) production measurement as described in Materials and methods. The values expressed in this table represent an average of triplicate determinations of three to five experiments with an SD of less than F5%. aminoglycan found in the extracellular matrix of mammalian tissues, is now considered to be both a physiologically relevant ligand and an adhesion molecule [38,39]. Modified HA has been used to promote wound healing and fibrovascular tissue growth [40]. HA conjugated to anticancer drugs can target HA receptor overexpressing tumor cells [41,42]. The receptors for HA, which include numerous hyaladherins such as CD44, RHAMM and aggrecan, regulate cellular responses to endogenous HA and chemically modified HA [43]. One of the major receptors for HA on the cell surface of aortic endothelial cells is CD44 [3 5]. CD44 is encoded by a single gene, which contains 19 exons (10). The most common form, CD44s (CD44 standard form), contains exons 1 5 (Nterminal 150 aa), exons 15 and 16 (membrane proximal 85 aa), exon 17 (transmembrane domain), and a part of exons 17 and 19 (cytoplasmic tail, 70 aa) [2]. Aortic endothelial cells (GM7372A cell line) contain an additional exon v10 which is inserted into the CD44s transcripts [4]. This isoform has been designated as CD44v10 (Figs. 2, 3 and 5), and is further modified by extensive N- and O-glycosylations and glycosaminoglycan (GAG) additions [4]. Apparently, both posttranslational modifications and/or alternative splicing within the CD44v10 structure determine the functional outcome of this molecule in endothelial cells. The binding of multivalent HA to CD44 is capable of inducing the clustering of many CD44 molecules. This in turn amplifies CD44 interactions with many different intracellular regulatory molecules leading to cellular signaling and biological activities [1]. Lipid rafts, also called caveolae, have been implicated in a variety of cellular functions including cholesterol and calcium regulation, and signal transduction [12,14]. In GM7372A cells, we have showed that CD44v10 is located in specialized caveolin-1-rich plasma membrane microdomains containing high cholesterol (designated as lipid rafts ) (Figs. 1 and 2). The binding of HA to GM7372A cells increases CD44v10 accumulation into the caveolin-1- containing lipid rafts (Figs. 1 and 2). This event is closely coupled with the onset of intracellular Ca 2+ mobilization (Fig. 6). Our observation that depletion of cholesterol by treating cells with methyl-h-cyclodextrin blocks HA-induced CD44v10 co-localization with caveolin-1 in lipid rafts (Fig. 2) and intracellular Ca 2+ mobilization (Fig. 6) suggests that cholesterol in lipid rafts directly influences HA/CD44-mediated Ca 2+ signaling in aortic endothelial cells. Fig. 6. Analyses of HA-induced Ca 2+ activity in GM7372A cells transfected with ARDcDNA or vector alone. (A) Measurement of 45 Ca 2+ flux in phospholipid vesicles (liposomes) containing IP 3 receptor. Isolated IP 3 receptor (obtained from lipid rafts of transfectants treated with HA or no HA) (25 Ag/ml) were incorporated into phosphatidylcholine/phosphatidylserine vesicles (liposomes). These IP 3 receptor subtype-containing phospholipid vesicles (liposomes) were then used for the measurement of IP 3 -induced 45 Ca 2+ influx as described in Materials and methods. (a) The amount of IP 3 -induced 45 Ca 2+ flux in vesicles containing IP 3 receptor from lipid rafts isolated from vector-transfected cells treated with no HA. (b) The amount of IP 3 -induced 45 Ca 2+ flux in vesicles containing IP 3 receptor from lipid rafts isolated from vector-transfected cells treated with HA. (c) The amount of IP 3 -induced 45 Ca 2+ flux in vesicles containing IP 3 receptor from lipid rafts isolated from ARDcDNA-transfected cells treated with no HA. (d) The amount of IP 3 -induced 45 Ca 2+ flux in vesicles containing IP 3 receptor from lipid rafts isolated from ARDcDNA-transfected cells treated with HA. (B) Measurement of HA-mediated Ca 2+ mobilization in GM7372A cells transfected with ARDcDNA or vector alone. Intracellular Ca 2+ mobilization was measured by a fluorescence spectrophotometer using cells [transfected with vector alone (a c) or ARDcDNA (d)] loaded with Fura-2/AM as described in Materials and methods. Subsequently, Fura-2-labeled transfectants were then treated with HA (indicated by arrowhead). (a) HA-mediated Ca 2+ mobilization in vector-transfected cells treated with no drug. (b) HA-mediated Ca 2+ mobilization in vector-transfected cells with Xestospongin C. (c) HA-mediated Ca 2+ mobilization in vector-transfected cells with methyl-h-cyclodextrin. (d) HA-mediated Ca 2+ mobilization in ARDcDNA-transfected cells treated with no drug. (B-b, c and d) The insert of (b, c and d) illustrates Ca 2+ release from internal stores by ionomycin and an inhibition of Ca 2+ elevation by EGTA (indicated by arrows) in Xestospongin C-treated (b) or methyl-h-cyclodextrin-treated (c) or ARDcDNA-transfected cells (d) preincubated with 1.2mM CaCl 2.